Machine Learning in E-Commerce: Trends, Applications, and Future Challenges

The rapid evolution of e-commerce has been significantly influenced by the integration of machine learning (ML) and data science techniques. The present survey provides a comprehensive overview of how ML methods are applied across various functional domains in e-commerce, including personalized recommendations, dynamic pricing, fraud detection, customer segmentation, and behavioral analysis. We categorize and evaluate a wide range of ML paradigms, namely supervised, unsupervised, reinforcement, and hybrid learning, as well as emerging approaches such as neurosymbolic artificial intelligence (AI), federated learning (FL), and quantum ML (QML). Key challenges related to scalability, interpretability, cold-start problems, data sparsity, and privacy are critically analyzed. Additionally, we highlight underexplored areas, such as continual learning (CL) and multi-agent architectures in commerce. The survey incorporates comparative tables, real-world use cases, and a taxonomy of methods to support both academic and industrial perspectives. Ultimately, by analyzing trends and gaps in the literature, we provide a forward-looking research roadmap that bridges ML innovations with the evolving demands of e-commerce ecosystems.